Although many studies have examined Early Childhood Caries (ECC), the etiology behind demonstrated racial/ethnic differences remain unknown. For almost two decades, researchers have tried to develop sensitive and specific caries risk assessment tools to identify at-risk individuals, but with limited success. This study will utilize statewide and local cross-sectional data sources to develop, test and refine an ECC risk association model, which will multifactorial and multilevel approach looking at various individual family and community characteristics, including demographics, socioeconomics, acculturation, health beliefs and behaviors, access to care, and physiology, using computationally intensive statistical data mining methods such as artificial neural networks. Aims are: to develop an ECC risk association model with individual- and family-level data from the 1993-4 statewide California Oral Health Needs Assessment supplemented with neighborhood-level data; to test and refine the above ECC risk association model and to test and refine th final ECC risk association model as a risk prediction model with the control group from a longitudinal fluoride varnish trial in San Francisco using individual, family, and neighborhood-level data. Risk assessment models to be used in non-dental settings and race/ethnicity specific models will be developed. This study interfaces with three already funded studies, as well as the other projects and the cores of the proposed center. The project will provide input to the San Ysidro intervention study, the San Francisco focus group, and the Kaiser Permanente health services research study about factors associated with ECC and provide risk assessment models that can be tested and refined with their study data. This is the first study to simultaneously examine individual-, family- and community-level characteristics related to ECC with data mining techniques.